Artificial intelligence is having a moment. In fact, Google Chief Executive Sundar Pichai just anointed AI as the future of computing. “Over time, the computer itself — whatever its form factor — will be an intelligent assistant helping you through your day,” he said in a recent letter to investors. “We will move from mobile first to an AI first world.”
Pichai added that artificial intelligence will eventually tackle even the world’s biggest challenges, such as cancer diagnosis. And a number of startups, particularly in the healthcare space, are taking those words to heart.
Venture-backed companies Sense.ly, AiCure and Babylon Health are leveraging AI to revolutionize healthcare. And healthcare-related AI startups accounted for 15 percent of all AI deals in 2015, according to research firm CB Insights.
Healthcare and AI “are extremely complementary,” says Rob Kniaz, founding partner of London-based Hoxton Ventures, which participated in a $25 million Series A round in Babylon Health. While doctors learn to pattern-match very well, AI can do a better job and won’t overlook any details, he says.
“AI systems can have a huge advantage in diagnosing those outlier cases,” Kniaz says. “When doctors are rushed, mistakes can be made. But AI is always very holistic and dispassionate in its analysis.”
Babylon, a health app that lets patients connect with medical professionals via mobile devices, recently put that thesis to the test. Babylon has a new AI-powered feature called “Check a Symptom” that asks users a series of questions and then instantly provides the most appropriate medical advice.
As a PR stunt, Babylon recently pitted its app against an Oxford-educated doctor and a nurse with 20 years of accident and emergency experience, and the app held its own. In fact, Babylon says its AI outperformed both doctors and nurses across 102 mock patient consultations.
Babylon currently operates in the United Kingdom, where patients often line up for hours to see their general practitioners. “AI can be hugely beneficial in relieving pressure from that system,” Kniaz says. “Being able to quickly triage those patients and offload the more basic cases is a huge win for everyone.”
Kniaz says Babylon has hit on a multibillion-dollar opportunity, but he admits that several challenges remain.
A key one is simply getting patients accustomed to engaging with an AI interface, which does not have the same bedside manner as a human doctor. But Babylon says patients might be more forthcoming in sharing embarrassing symptoms, such as a herpes outbreak, with a computer, which does not have the ability to pass moral judgment.
Though it is still early days, AI could eventually have a very high payoff in healthcare. A recent study by research firm Frost & Sullivan found that AI has the potential to improve health outcomes by 30 percent to 40 percent while cutting treatment costs by as much as 50 percent.
Frost & Sullivan estimates the market for AI and cognitive computing systems in healthcare will reach $6.7 billion in 2021, more than 10 times the $634 million of 2014.
“Already playing a critical role in other industries, AI systems are poised to transform how we think about disease diagnosis and treatment,” the firm noted in its report.
Anis Uzzaman, general partner and chief executive of Fenox Venture Capital, is one of the most active investors in AI. He says the healthcare-AI convergence can help democratize the delivery of medical care, especially in the developing world where doctors and nurses are in severe short supply.
That’s one of the reasons his firm recently led a $2.2 million round in Sense.ly, a telemedicine platform that leverages AI to provide users with an interactive virtual nurse called Molly.
Using Sense.ly’s smartphone application, a patient can interact with Molly using simple voice commands to ask a wide range of health-related questions. The AI technology is used to understand the meaning of the spoken questions and provide answers or relevant medical information.
Sense.ly maintains that since many health situations can be addressed without a doctor or nurse consultation, its technology promises high-quality care at reduced cost.
“In the past you used to go to the bank to deposit money, but now you can just use PayPal or mobile deposit,” Uzzaman says. “The same thing is happening in the healthcare industry.
“With AI, you won’t have to go the hospital to get a diagnosis, which can reduce our dependence on doctors and nurses, who are already in short supply, especially in the developing world. Essentially, we are trying to solve this problem of providing quality care to everyone.”
Soon, says Uzzaman, healthcare applications won’t just have the ability to think; they will also be able to feel.
His firm recently led a $40 million round in Affectiva, a specialist in emotion-recognition software and analysis. Basically, the technology can read facial expressions through any device equipped with a camera and can understand how a person is feeling. He says a company like Sense.ly could eventually integrate this technology into its own platform.
“The app would be able to read your facial expression and immediately detect if you are tired or stressed or ill. It could then quickly alert a healthcare professional if needed,” he says.
AI is starting to find its way into many niches within the healthcare industry. Take, for instance, clinical trials for new drugs. A major problem with drug testing is that some patients in trials do not actually take the prescribed medications. That leads to bad data, and bad data leads to problems with the approval process for new drugs.
“Imagine if a pharmaceutical company could shrink the time of its clinical trial by having 100 percent accurate data,” says Brian Hirsch, co-founder and managing partner at Tribeca Venture Partners. “How meaningful is it to speed the approval process and get a drug to market months or years faster? The difference could amount to billions of dollars.”
That’s why he invested in AiCure, which raised more than $12 million in a Series A round earlier this year. The company’s smartphone app uses facial-recognition and motion-sensing software to observe patients in clinical trials and detect whether they are taking their medications.
The app uses AI and deep learning to figure out whether patients are swallowing the pill or are just sticking it in a cheek and then spitting it out. The system directly visualizes patient behavior with no human review necessary. Alerts are sent out if the app detects non-adherence or suspicious behavior.
“This is a fascinating application of AI to solve a real-world problem,” Hirsch says. “This is not just AI for AI’s sake.”
He adds that AiCure has other potential applications. For instance, insurers could use the application to monitor whether someone with high cholesterol is actually taking Lipitor. Hirsch says the insurer could even provide the drug free, as long as patients use an app like AiCure to verify they are taking it, thus reducing the risk of expensive hospitalizations.
“Verifying medication intake as a way to improve health outcomes and lower costs can be very powerful for insurers,” he says.
What’s also powerful is the exit potential of AI-centric healthcare startups.
“Just look at what Google’s chief executive said about AI,” notes Uzzaman. “All the big technology players want a piece of AI because they cannot lead without it. They simply won’t allow a promising AI startup to reach the IPO stage. They will make an attractive offer long before that.”
Tom Stein is a Palo Alto, California-based contributor. He can be reached at firstname.lastname@example.org.
Photo of online medical consultation concept on tablet screen courtesy of ©iStock.com/CarmenMurillo